"blather" - izing livejournal
Jul. 19th, 2004 10:42 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
saint_monkeyI used to love the website blather. The concept was (and is, i guess,) cool. you put an entry in, and the system sorted it and linked each word in the entry to titled entries containing the same word. ie: you write the word "jesus" in an entry about frat boys, and the word "jesus" would automatically become a hyperlink to the entry about "jesus." When i found that i could pull and parse the latest.bml entries from livejournal, I began writing a tool that could "Blatherize" a journal entry... it groups words in sets of five, searches for the phrase, then in sets of four, etc, down to one. if it finds a match, say "nothing better to do" in any of the latest posts, then it makes a link to the entry. here it is at work on dagoski's latest post:
I'm a little bored today. I'm a lameduck at work as I go into the final four here. So, I'm coming up with yet another geek project. I was just looking at my friends and friend of lists and noting that they don't quite have a one to one correspondance. That got me thinking a little about how people are connected in Live Journal and other similar online networks(like ICQ etc...). What I'm looking for are the sites of nucleation for social networks. Can such be found by analysing the topology represented in these lists? And is it possible to infer RL relationships as opposed to purely online ones? I'm sure someone's done this before, but, like I said, I'm bored. And, once upon a time, I wondered about the thermodynamics of large computer networks, did a lot of data collection and a fair amount of analysis on my night shift at the Network Ops Center. I shelved the project when I was tapped to be an assistant system administrator, meaning that I ceased to have nothing better to do on the night shift. Two years later, what do I spy in Nature? Yep, a paper on the thermodynamics of large computer networks. Some problems just seem to have a time of their own and many people start looking at them at once. The person who winds up credited is just the guy(or gal) who passes peer review first.
Given that I'm bored and nothing really matters much at work, what problems do y'all think I should ponder for the next four days? I mean, aside from getting a new job and applying for unemployment?
of course, there are lots of limitations. it'd be nice to be able to parse a larger set of entries, then we could assuredly match more than just prepositional phrases and the like, but i'm limited to what the feed provides, and thing is hardly instant. it takes five minutes to run on a mac, and a minute on PC, and according to iPulse, eats about 82% of the processor, so I won't bore folks with the code unless they want to see it out of morbid curiosity.
dagoski's latest post:I'm a little bored today. I'm a lameduck at work as I go into the final four here. So, I'm coming up with yet another geek project. I was just looking at my friends and friend of lists and noting that they don't quite have a one to one correspondance. That got me thinking a little about how people are connected in Live Journal and other similar online networks(like ICQ etc...). What I'm looking for are the sites of nucleation for social networks. Can such be found by analysing the topology represented in these lists? And is it possible to infer RL relationships as opposed to purely online ones? I'm sure someone's done this before, but, like I said, I'm bored. And, once upon a time, I wondered about the thermodynamics of large computer networks, did a lot of data collection and a fair amount of analysis on my night shift at the Network Ops Center. I shelved the project when I was tapped to be an assistant system administrator, meaning that I ceased to have nothing better to do on the night shift. Two years later, what do I spy in Nature? Yep, a paper on the thermodynamics of large computer networks. Some problems just seem to have a time of their own and many people start looking at them at once. The person who winds up credited is just the guy(or gal) who passes peer review first.
Given that I'm bored and nothing really matters much at work, what problems do y'all think I should ponder for the next four days? I mean, aside from getting a new job and applying for unemployment?
of course, there are lots of limitations. it'd be nice to be able to parse a larger set of entries, then we could assuredly match more than just prepositional phrases and the like, but i'm limited to what the feed provides, and thing is hardly instant. it takes five minutes to run on a mac, and a minute on PC, and according to iPulse, eats about 82% of the processor, so I won't bore folks with the code unless they want to see it out of morbid curiosity.
